examifyr

Readiness test · CompTIA · Free · No sign-up required

CompTIA SecAI+ Readiness Test — Are You Ready for CY0-001?

SecAI+ launched on 17 February 2026 as the first vendor-neutral certification focused on securing AI systems and using AI in defence. This free readiness test samples questions across all four exam domains — weighted by CompTIA’s published blueprint — and tells you where you stand before you book.

~30 questions sampled across every exam domain, weighted by the official blueprint · no sign-up

Instant resultsTopic breakdownWeak area detectionPass likelihood

What this test covers

SecAI+ (exam code CY0-001) targets security practitioners who have to secure AI systems rather than merely use them, and its weighting says so plainly: Securing AI systems alone is 40% of the exam, covering adversarial risk, securing AI deployments on-premises and in cloud, and the controls around models and the pipelines feeding them. AI-assisted security is a further 24%, AI governance, risk and compliance 19%, and basic AI concepts for security 17%. Unlike a flat practice quiz, this readiness test mirrors that blueprint: questions are drawn across the four domains in proportion to their published weight, so your score reflects what the exam actually emphasises. When you finish you get a readiness score from 0-100, a pass-likelihood band, and a per-domain breakdown naming your weakest areas — for example "weak in Securing AI systems, 40% of the exam" — so you know precisely what to study next. Fresh questions every attempt. No sign-up, no email required.

  • Securing AI systems — 40%
  • AI-assisted security — 24%
  • AI governance, risk, and compliance — 19%
  • Basic AI concepts related to cybersecurity — 17%

How it works

01

Choose a level

Easy, Medium, or Hard for the full readiness test. Every level is sampled across the exam domains in proportion to their published weight.

02

Answer questions

One question at a time. Click an answer and get instant feedback with a full explanation.

03

Get your score

See your readiness score (0–100), pass likelihood, and a per-domain breakdown of your weakest exam areas.

What the questions look like

6 worked examples, with the reasoning spelled out. These are written for this page rather than taken from the live bank — showing a real question would spend it as a measurement for anyone who read this first.

Securing AI systems — 40%

An LLM assistant summarises a user's inbound email. An attacker sends a message containing hidden text instructing the assistant to forward the whole mailbox to an outside address. What is this attack?

  1. AModel inversion
  2. BTraining data poisoning
  3. CIndirect prompt injectionCorrect
  4. DMembership inference
WhyThe malicious instructions arrive inside content the model processes on the user's behalf rather than from the user, which is what makes the injection indirect. Model inversion and membership inference extract information about training data, and poisoning tampers with training data before the model is built.

Securing AI systems — 40%

For the email assistant above, which control most limits the damage if an injection succeeds?

  1. ALeast-privilege tools, with human approval before it can send mailCorrect
  2. BA system prompt instructing the model to ignore any instructions found inside emails
  3. CRetraining the model on examples of injected emails so that it learns to recognise them
  4. DEncrypting every stored email at rest so that the mailbox contents stay protected
WhyInjection cannot be reliably prevented by prompting or training, so the control that matters limits what a manipulated model is able to do. If the assistant cannot send mail without a person approving it, a successful injection has nothing dangerous to trigger. Encryption at rest does not apply, because the assistant reads the mail decrypted.

Securing AI systems — 40%

A team fine-tunes a model on text scraped from public forums. An adversary had posted crafted content so the model learns to behave maliciously whenever a specific trigger phrase appears. What is this?

  1. APrompt injection
  2. BData poisoning with a backdoor triggerCorrect
  3. CModel extraction
  4. DDenial of wallet
WhyThe attack happens through the training data, before the model is ever deployed, and leaves hidden behaviour tied to a trigger — a poisoning backdoor. Prompt injection happens at inference time, extraction copies a model through its outputs, and denial of wallet drives up usage costs.

AI-assisted security — 24%

A security operations team adds an LLM to help triage alerts. What is the most appropriate way to use its output?

  1. AAs final verdicts that automatically close any alert the model labels benign
  2. BAs a replacement for the correlation rules that already run in the SIEM
  3. CAs auto-remediation for any alert the model rates as high confidence, without review
  4. DAs a prioritised recommendation an analyst verifies before actingCorrect
WhyLLM output is useful for speeding up triage but can be wrong and confidently so, which makes it decision support rather than a decision. Automatic closure or remediation lets one model error become a missed intrusion or an outage, and deterministic correlation rules do a different job the model does not replace.

AI governance, risk, and compliance — 19%

Under the EU AI Act, how is an AI system used to screen job applicants' CVs classified?

  1. AHigh-riskCorrect
  2. BProhibited
  3. CMinimal risk
  4. DLimited risk, with transparency obligations only
WhyThe Act lists AI used in employment and recruitment — including screening and filtering applications — among its high-risk uses, which brings obligations such as risk management, data governance, human oversight and logging. Prohibited practices are a separate, narrower list, and minimal or limited risk would understate a system that shapes who gets hired.

Basic AI concepts related to cybersecurity — 17%

Why can a large language model state an incorrect fact with apparent confidence?

  1. AIts training data is stored encrypted, so parts of it cannot be read back
  2. BIt deliberately hides uncertainty so that it appears more helpful to users
  3. CIt generates likely sequences of tokens rather than looking up verified factsCorrect
  4. DIts temperature is fixed at zero, which removes any doubt from its answers
WhyA language model predicts plausible continuations of text. Fluency and correctness are different properties, so an answer can read as certain while being wrong — the behaviour usually called hallucination. It has no intent to hide anything, and temperature is a setting rather than something fixed.

Frequently asked questions

What is the CompTIA SecAI+ exam?
CompTIA SecAI+ (exam code CY0-001) launched on 17 February 2026. It is a vendor-neutral certification covering the security of AI systems — securing AI deployments and mitigating adversarial risk, AI-assisted detection and response, AI governance, risk and compliance, and the AI concepts a security practitioner needs. The exam runs 60 minutes with a maximum of 60 questions, mixing multiple-choice and performance-based items.
What is the passing score for SecAI+?
CompTIA requires 600 on a scale of 100 to 900. That is a scaled score, not a percentage of questions answered correctly, so it cannot be converted into "I need N right" — which is also why a readiness score is best read as a per-domain diagnosis rather than a predicted result.
How hard is SecAI+, and what experience does it assume?
CompTIA recommends 3-4 years in IT with 2 or more years hands-on in cybersecurity, plus Security+, CySA+, PenTest+ or equivalent knowledge. It is not entry-level: it assumes you already work in security and need to extend that into AI systems. The 40% Securing AI systems domain is where candidates from a pure security background most often find gaps, because it asks about model and pipeline controls rather than network and endpoint ones.
Do I need Security+ before taking SecAI+?
It is recommended rather than required. CompTIA names Security+, CySA+ or PenTest+ as equivalent background, and the exam assumes that baseline instead of teaching it — foundational security concepts appear as context for AI-specific questions rather than as the subject of them.
How is this different from the Cybersecurity practice test?
It draws on a different question bank. This test scores you against CompTIA’s four SecAI+ domains at their published weights — adversarial attacks, securing AI deployments, AI-assisted detection, AI governance — whereas the Cybersecurity practice test covers general security fundamentals. If you want a SecAI+ readiness score, this is the one that measures it.
What does the readiness score mean?
A number from 0-100 with a band: 85-100 Exam Ready, 70-84 Almost Ready, 50-69 At Risk, 0-49 Not Ready. Each band carries a pass-likelihood estimate and a per-domain accuracy breakdown. Because CompTIA scores the real exam on a 100-900 scale, aim comfortably clear of the line rather than at it.
Is it free? Do I need to sign up?
Completely free, no account needed. Pick the test, answer the questions, and get your domain-weighted readiness score instantly.

Related